The answer desk

Why was Dholera chosen for the fab?

Published 25 August 2026. Facts verified to 25 August 2026 unless dated otherwise.

No public document sets out why Dholera was chosen, so the honest answer is read from structure rather than narrative. Dholera could supply what a wafer fab needs and a metropolitan technology hub cannot: hundreds of contiguous serviced acres, a notified Special Economic Zone, bulk power already commissioned, and a state incentive package written with Dholera Semicon City named inside it. Bangalore's semiconductor strength is design and equipment engineering, which is a different layer of the industry. No wafer fab has been approved for Karnataka under the India Semiconductor Mission.

The short answer, and the part that has never been published

Start with what is missing, because the missing thing shapes every honest answer to this question. There is no published site-selection study for the Dholera fab. No Cabinet note, company filing or ministry release in the public record ranks candidate sites, scores them against criteria, or explains what Gujarat offered that another state did not. Anyone who tells you the decision came down to one clean factor is reconstructing a story from an outcome. This desk will not do that.

What is public is the structure of the offer, and structure is more informative than narrative anyway. By the time the Union Cabinet approved the plant on 29 February 2024 at Rs 91,000 crore DURABLE, Gujarat had spent more than a decade building the one thing a wafer fab cannot improvise, which is a large contiguous parcel of serviced industrial land with a single government counterparty, and had already written a state incentive policy that names Dholera inside its text. The fab did not arrive and then discover a suitable site. The site had been manufactured first, and the fab landed on it.

The Bangalore half of the question rests on a category error worth correcting early. Bangalore is where the largest share of India's chip design and verification work sits, alongside the India engineering operations of several global toolmakers. None of that work needs a wafer fab, and a wafer fab does not need any of it in the same city. The two belong to different layers of the same industry, and India's approved semiconductor manufacturing projects have gone to Gujarat, Assam, Uttar Pradesh, Odisha, Punjab and Andhra Pradesh rather than to Karnataka.

What a wafer fab actually demands from a site

Before any comparison means anything, the requirement list has to be concrete. A 300mm fab is not a large factory with tighter cleaning standards. It is a chemical plant, a power substation, a water treatment works and a vibration-isolated precision laboratory stacked on one slab, and its site requirements follow from that.

• Contiguous land, at campus scale, with room to double. A fab occupies its own plot and then needs adjacent ground for gas yards, chemical stores, waste treatment, substations, contractor laydown and a second phase. Fragmented parcels assembled from many owners do not work, because a single holdout can stall commissioning.
• Ground that can be made quiet. Lithography tools are specified against vibration criteria measured in microns per second. Whatever the soil is, the foundation has to deliver that, and the cost of getting there varies enormously by site.
• Bulk water, continuously, at semiconductor purity after treatment. Ultrapure water is made on site, but the raw feed has to be reliable at industrial volume, and no fab can share a municipal supply that rations in a dry year.
• Firm power with clean quality. Not simply megawatts. Sags and interruptions scrap work in progress across an entire line.
• Bulk gases produced over the fence. Nitrogen at fab volumes is not economically trucked, which is why an air separation unit tends to be built beside the fab rather than supplied to it.
• Chemical logistics and a customs regime. Almost every tool, precursor, photoresist and spare part is imported, so duty treatment and freight access are direct line items in the cost of production.

Read that list once and the siting question stops being about prestige addresses. It becomes a question about which authority can hand over serviced ground with utilities attached, and which state will underwrite the recurring inputs.

Dholera's real product: land manufactured, not assembled

Most Indian industrial projects fail during land assembly rather than during construction, which is exactly the failure mode Dholera's machinery was designed to remove. Land inside the Special Investment Region is supplied through town planning schemes, a Gujarat instrument that pools holdings, reserves a share for roads and infrastructure, and returns reconstituted final plots to the original owners with development rights attached, instead of buying owners out one at a time. Our land system guide sets out the mechanism in full.

The delivery record puts numbers on it. The NICDC and DPIIT Delivery Monitoring Unit report dated 31 May 2026 records that the Gujarat government transferred 48.31 square kilometres to the project, that trunk infrastructure works across the 22.5 square kilometre Phase I activation area are complete, and that 12 plots had been allotted, of which 436 acres are industrial, with Tata Chemicals named as anchor investor DURABLE. The same report describes industrial and commercial land as readily available for allotment. The institutional counterparty has existed since 28 January 2016, when DICDL was formed with Gujarat holding 51 percent through DSIRDA and the Centre 49 percent through the NICDC Trust.

That combination is the actual asset. A promoter negotiating for a fab site at Dholera deals with one government special purpose vehicle holding transferred land inside a sanctioned scheme, and receives a plot with defined access and utility entitlements. Nothing on the public record shows a metropolitan alternative in India offering the equivalent at that scale, and the reason is not administrative willingness. It is that cities have already been built on the land.

The state wrote its incentive with one address inside it

Here the record gets unusually explicit, and it is the closest thing to a documented siting rationale that exists. The Gujarat Semiconductor Policy 2022 to 2027, announced on 27 July 2022, was the first state semiconductor policy in India. Its terms are not generic. The state contributes 40 percent of the capital assistance approved by the Centre, and offers a land subsidy of 75 percent on the first 200 acres for a fab in Dholera Semicon City with 50 percent on additional land, water at Rs 12 per cubic metre for five years rising ten percent annually thereafter, a power subsidy of Rs 2 per unit for ten years with electricity duty exemption, and a 50 percent capital subsidy for a desalination plant built within the first five years DURABLE. The line-by-line reading sits on our incentives file.

Two things follow. First, a fab located elsewhere in Gujarat would not have collected the same land term, because the land subsidy is written against Dholera Semicon City specifically. Second, the package attacks the two costs that decide whether a mature-node fab stays competitive after commissioning, which are water and electricity, rather than only the capital cost that subsidy schemes usually target. A state that subsidises the inputs a plant consumes for thirty years has thought about the plant rather than the announcement.

The customs layer arrived later and completed the picture. A Special Economic Zone of 66.16 hectares was notified by the Ministry of Commerce and Industry for Tata Semiconductor Manufacturing Private Limited around 16 April 2026. SEZ status matters to a fab mainly because the tool set is imported, and duty-free import of capital goods changes the landed cost of equipment measured in billions of dollars. Our SEZ guide explains what the status does and does not confer, and the short version of the limit is that it decides duty treatment and decides nothing about yield or customers.

The utilities case, scored honestly

Siting arguments usually collapse into brochure claims about infrastructure. The discipline here is to separate what is commissioned from what is tendered.

Power is the strong column. Of the 1,000 MW sanctioned for Phase I of the Dholera solar park, about 300 MW awarded to Tata Power Solar is commissioned DURABLE, with roughly 700 MW under development by ReNew, SJVN, Vena and TEQ Green targeting March 2027. A commissioned generating asset is inspectable, which is why this record weights it more heavily than any memorandum, and the solar park sits beside the fab rather than at the far end of a transmission corridor.

Water is the weak column, and pretending otherwise would be the easiest way to fail this question. A tender was issued on 9 July 2026 to appoint consultants for a 200 MLD seawater desalination plant REPORTED, set against a current facility of around 20 MLD. A consultant appointment is not a plant, and the gap between the two is measured in years. Our water file keeps the running record. Anyone arguing that Dholera was selected because its water position was solved has the sequence backwards: the water solution is being built after the fab was sited, which is a risk the project is carrying rather than an advantage it enjoyed.

Connectivity sits between the two. The roughly 109 kilometre Ahmedabad to Dholera expressway on NH-751 has been open since 31 March 2026. The first trial landing at the Dholera airport took place on 4 June 2026, with operations targeted for September to October 2026 TARGET. The semi high speed Ahmedabad to Dholera rail link, valued at Rs 20,667 crore, is cleared with a target of 2030 to 2031. Road delivered, air imminent, rail distant: that ordering matters for a plant that will import tools by air freight and ship chemicals by road.

What the ground cost, which nobody advertises

The single most clarifying fact about Dholera's selection is that the ground was not ideal. The Economic Times reported on 28 November 2025 that Fugro, Cengrs and Geo Dynamics were engaged to redesign the main fab foundations after soil testing found the ground unsuitable as originally designed REPORTED. Our foundation file carries the investigation.

The Bhal is coastal and low-lying, and delivering single-digit micron per second vibration performance on ground that testing has flagged as unsuitable normally means deep piling, ground improvement, or both. That is expensive, and it lands on the critical path because nothing above the foundation can proceed until it is resolved. The correct inference is not that the site was a mistake. It is that whoever weighed the options accepted a known geotechnical cost in exchange for land supply, utilities and a fiscal package. Selection criteria are revealed by what the selector was willing to pay for, and here the record shows the buyer paying in foundations to obtain everything else.

Why Bangalore was never the comparison it sounds like

Karnataka's position in semiconductors is real and is not fab-shaped. Bangalore hosts design services houses, captive design centres of global chip companies, verification and physical design teams, and the India engineering operations of several equipment makers, and this record keeps the dated value of each company's India commitment on its company files rather than asserting it here. None of that work produces a patterned wafer, and none of it requires 200 acres of serviced ground next to an air separation unit. A design centre needs engineers, licences and bandwidth. A fab needs land, water, power and a customs regime, and the two lists barely intersect.

The clean test is the approvals register. Across the India Semiconductor Mission's portfolio, manufacturing units have been sanctioned in Gujarat, Assam, Uttar Pradesh, Odisha, Punjab and Andhra Pradesh, including the four units approved on 12 August 2025 for Odisha, Punjab and Andhra Pradesh with a cumulative investment of about Rs 4,600 crore DURABLE. Karnataka does not appear in that manufacturing register, and that absence is a fact about what fabs need rather than a judgement about the state.

Read the rows together and the answer resolves itself. Four of the five requirements point away from a large city, and the fifth points towards one. A country building its first commercial fab optimises for the four it cannot buy later, and imports the fifth by flying engineers to the partner's fabs. That is what has been happening: the Economic Times reported in June 2025 that Tata had sent a couple of hundred people to PSMC in Taiwan for training, in batches of around seventy-five across equipment, yield, process technology and quality engineering REPORTED. Talent is the one input on the list that can be moved by aircraft.

The comparison that actually happened was between states, not cities

The competition for these plants runs through state incentive packages rather than through municipal amenities, because the central scheme covers a fixed share of eligible cost and the state top-up is the variable. Gujarat published first, in July 2022, and published terms specific enough to price. That is a structural advantage independent of geography, and it is why the same state now holds the front end at Dholera and the back end at Sanand, a pairing our siting comparison reads in detail.

The honest caveat is that a fiscal auction between states is not the same thing as an engineering site selection, and the public record does not let anyone say which weighed more. What can be said is that the two pointed the same way in this case, which removes the interesting version of the question. Had the best fiscal offer come from a state without serviced land, the decision would have been genuinely difficult and the outcome genuinely informative. It did not.

What is not public, and what to watch next

Four things would convert this answer from structural inference into documented fact, and none of them exists in the public record as of 25 August 2026. A company statement describing the site selection criteria and the shortlist. A ministry document recording how proposals were scored. Any statement from Tata Electronics or PSMC ranking candidate locations. And the terms of the Fiscal Support Agreement signed on 5 March 2025, which have never been published even though the agreement's existence and date are on the record.

Until those arrive, this desk reports the structure and refuses the story. The structure is strong enough on its own: land that had been assembled through a public instrument for a decade, a state policy naming the location, a notified customs zone, commissioned power, an unsolved water programme, and a foundation bill the promoter accepted rather than avoided.

Three observable events would test whether the siting logic holds. The first is expansion behaviour: the Union Cabinet cleared two further semiconductor units for Gujarat in May 2026, at Dholera and Surat, with a cumulative investment stated at more than Rs 3,900 crore DURABLE. Repeat investment at the same address is the market's own verdict on the site. The second is the water programme moving from consultants to construction, because a fab at volume cannot run on a 20 MLD facility. The third is the allotment register: 12 plots is a small number, and suppliers taking ground near the anchor is how a site stops being a policy decision and becomes a cluster. Each of those is dated, checkable, and recorded here as it happens.